Scheduler学习之零：Kernel cpupri研究

最近开始详细研究scheduler中的RT调度器，其中有看到一个cpupri，作为选核的一部分，有必要先研究一下这个。

Summary:

为了方便描述：做如下的概念约定：

把一个cpu中正在排队以及正在running的所有rt task最高的优先级叫做此cpu的优先级。

根据这个优先级可以对CPU进行排序。
对于RT调度器来讲，其重要任务是保证优先级高的task优先执行，优先级高的task的响应时间也要最快。
假定一个task上面已经有比较高优先级的Task在排队或者run的话，这时候再摆上去一个优先级比较低的task，那么这个task 就会排队，造成runnable，即便是其他cpu上面所有task的优先级都比这个task还低。这样显然不合理。

而cpupri则负责跟踪记录各个cpu中所有task的最高优先级情况。方便后面寻找优先级最低的cpu出来，供RT调度器进行选择。

代码分析

其位于：

kernel/sched/cpupri.h，内容如下所示。

/* SPDX-License-Identifier: GPL-2.0 */

#define CPUPRI_NR_PRIORITIES	(MAX_RT_PRIO + 2)

#define CPUPRI_INVALID		-1
#define CPUPRI_IDLE		 0
#define CPUPRI_NORMAL		 1
/* values 2-101 are RT priorities 0-99 */

struct cpupri_vec {
	atomic_t		count;
	cpumask_var_t		mask;
};

struct cpupri {
	struct cpupri_vec	pri_to_cpu[CPUPRI_NR_PRIORITIES];//有多少个RT优先级就有多少个元素。每个元素记录吧这个优先级分别是哪些个cpu的最高优先级。所以理论上讲最多只有8个元素是有值的，假定有8个cpu的话。
	int			*cpu_to_pri;//一个有多少个cpu就有多少个元素的数组。每个元素代表对应cpu中当前最高优先级的值。
};

#ifdef CONFIG_SMP
int  cpupri_find(struct cpupri *cp, struct task_struct *p, struct cpumask *lowest_mask);
void cpupri_set(struct cpupri *cp, int cpu, int pri);
int  cpupri_init(struct cpupri *cp);
void cpupri_cleanup(struct cpupri *cp);
#endif

1. 使用cpupri的地方

如下所示在root_domain中定义了一个结构体对象。所以看上去属于全局变量。

/*
 * We add the notion of a root-domain which will be used to define per-domain
 * variables. Each exclusive cpuset essentially defines an island domain by
 * fully partitioning the member CPUs from any other cpuset. Whenever a new
 * exclusive cpuset is created, we also create and attach a new root-domain
 * object.
 *
 */
struct root_domain {
	………………………………
	/*
	 * The "RT overload" flag: it gets set if a CPU has more than
	 * one runnable RT task.
	 */
	cpumask_var_t		rto_mask;
	struct cpupri		cpupri;

	unsigned long		max_cpu_capacity;

    ……………………
};

在init_rootdomain函数中对其进行了初始化。

static int init_rootdomain(struct root_domain *rd)
{
	if (!zalloc_cpumask_var(&rd->span, GFP_KERNEL))
		goto out;
	if (!zalloc_cpumask_var(&rd->online, GFP_KERNEL))
		goto free_span;
	if (!zalloc_cpumask_var(&rd->dlo_mask, GFP_KERNEL))
		goto free_online;
	if (!zalloc_cpumask_var(&rd->rto_mask, GFP_KERNEL))
		goto free_dlo_mask;

#ifdef HAVE_RT_PUSH_IPI
	rd->rto_cpu = -1;
	raw_spin_lock_init(&rd->rto_lock);
	init_irq_work(&rd->rto_push_work, rto_push_irq_work_func);
#endif

	init_dl_bw(&rd->dl_bw);
	if (cpudl_init(&rd->cpudl) != 0)
		goto free_rto_mask;

	if (cpupri_init(&rd->cpupri) != 0)
		goto free_cpudl;
	return 0;

free_cpudl:
	cpudl_cleanup(&rd->cpudl);
free_rto_mask:
	free_cpumask_var(rd->rto_mask);
free_dlo_mask:
	free_cpumask_var(rd->dlo_mask);
free_online:
	free_cpumask_var(rd->online);
free_span:
	free_cpumask_var(rd->span);
out:
	return -ENOMEM;
}

/**
 * cpupri_init - initialize the cpupri structure
 * @cp: The cpupri context
 *
 * Return: -ENOMEM on memory allocation failure.
 */
int cpupri_init(struct cpupri *cp)
{
	int i;

	for (i = 0; i < CPUPRI_NR_PRIORITIES; i++) {
		struct cpupri_vec *vec = &cp->pri_to_cpu[i];//如前面pri_to_cpu定义所示，每一个优先级都对应一个cpupri_vec.

		atomic_set(&vec->count, 0);//设置计数器为0
		if (!zalloc_cpumask_var(&vec->mask, GFP_KERNEL))
			goto cleanup;
	}

	cp->cpu_to_pri = kcalloc(nr_cpu_ids, sizeof(int), GFP_KERNEL);//cpu_to_pri数组，其元素数量为cpu的数量。
	if (!cp->cpu_to_pri)
		goto cleanup;

	for_each_possible_cpu(i)
		cp->cpu_to_pri[i] = CPUPRI_INVALID;

	return 0;

cleanup:
	for (i--; i >= 0; i--)
		free_cpumask_var(cp->pri_to_cpu[i].mask);
	return -ENOMEM;
}

rq，rd,cpupri关系如下图所示：


对cpupri对象的修改主要通过cpupri_set完成，调用的地方如下所示：

/**
 * cpupri_set - update the CPU priority setting
 * @cp: The cpupri context
 * @cpu: The target CPU
 * @newpri: The priority (INVALID-RT99) to assign to this CPU
 *
 * Note: Assumes cpu_rq(cpu)->lock is locked
 *
 * Returns: (void)
 */
void cpupri_set(struct cpupri *cp, int cpu, int newpri)
{
	int *currpri = &cp->cpu_to_pri[cpu];
	int oldpri = *currpri;
	int do_mb = 0;

	newpri = convert_prio(newpri);

	BUG_ON(newpri >= CPUPRI_NR_PRIORITIES);

	if (newpri == oldpri)
		return;

	/*
	 * If the CPU was currently mapped to a different value, we
	 * need to map it to the new value then remove the old value.
	 * Note, we must add the new value first, otherwise we risk the
	 * cpu being missed by the priority loop in cpupri_find.
	 */
    /*
     *如上面的注释所示，先设定新priority的cpupri_vec
     */
	if (likely(newpri != CPUPRI_INVALID)) {
		struct cpupri_vec *vec = &cp->pri_to_cpu[newpri];

		cpumask_set_cpu(cpu, vec->mask);//设定其mask，表明对应的cpu上面有Priority的task。
		/*
		 * When adding a new vector, we update the mask first,
		 * do a write memory barrier, and then update the count, to
		 * make sure the vector is visible when count is set.
		 */
		smp_mb__before_atomic();
		atomic_inc(&(vec)->count);
		do_mb = 1;
	}
	if (likely(oldpri != CPUPRI_INVALID)) {
		struct cpupri_vec *vec  = &cp->pri_to_cpu[oldpri];

		/*
		 * Because the order of modification of the vec->count
		 * is important, we must make sure that the update
		 * of the new prio is seen before we decrement the
		 * old prio. This makes sure that the loop sees
		 * one or the other when we raise the priority of
		 * the run queue. We don't care about when we lower the
		 * priority, as that will trigger an rt pull anyway.
		 *
		 * We only need to do a memory barrier if we updated
		 * the new priority vec.
		 */
		if (do_mb)
			smp_mb__after_atomic();

		/*
		 * When removing from the vector, we decrement the counter first
		 * do a memory barrier and then clear the mask.
		 */
		atomic_dec(&(vec)->count);
		smp_mb__after_atomic();
		cpumask_clear_cpu(cpu, vec->mask);
	}

	*currpri = newpri;
}

其完整的调用关系如下图所示：

所以可以看到基本上是在enqueu/dequeu entity的时候更新cpupri的值。主要沿着enqueue_entiry这条路径看以下：

static void __enqueue_rt_entity(struct sched_rt_entity *rt_se, unsigned int flags)
{
	struct rt_rq *rt_rq = rt_rq_of_se(rt_se);
	struct rt_prio_array *array = &rt_rq->active;
	struct rt_rq *group_rq = group_rt_rq(rt_se);
	struct list_head *queue = array->queue + rt_se_prio(rt_se);

	/*
	 * Don't enqueue the group if its throttled, or when empty.
	 * The latter is a consequence of the former when a child group
	 * get throttled and the current group doesn't have any other
	 * active members.
	 */
	if (group_rq && (rt_rq_throttled(group_rq) || !group_rq->rt_nr_running)) {
		if (rt_se->on_list)
			__delist_rt_entity(rt_se, array);
		return;
	}

	if (move_entity(flags)) {
		WARN_ON_ONCE(rt_se->on_list);
		if (flags & ENQUEUE_HEAD)
			list_add(&rt_se->run_list, queue);
		else
			list_add_tail(&rt_se->run_list, queue);

		__set_bit(rt_se_prio(rt_se), array->bitmap);
		rt_se->on_list = 1;
	}
	rt_se->on_rq = 1;
/*前面完成task的入队动作之后，开始更新队列的最高优先级。*/
	inc_rt_tasks(rt_se, rt_rq);
}
static inline
void inc_rt_tasks(struct sched_rt_entity *rt_se, struct rt_rq *rt_rq)
{
	int prio = rt_se_prio(rt_se);

	WARN_ON(!rt_prio(prio));
//记录了rt entity的数量
	rt_rq->rt_nr_running += rt_se_nr_running(rt_se);
	rt_rq->rr_nr_running += rt_se_rr_nr_running(rt_se);
//如下更新rt queue中最高优先级
	inc_rt_prio(rt_rq, prio);
	inc_rt_migration(rt_se, rt_rq);
	inc_rt_group(rt_se, rt_rq);
}

static void
inc_rt_prio(struct rt_rq *rt_rq, int prio)
{
	int prev_prio = rt_rq->highest_prio.curr;

	if (prio < prev_prio)
		rt_rq->highest_prio.curr = prio; //所以higest_pri.curr记录了当前队列中优先级最高的priority.由于rt qeueue当中，task是按优先级放人，相同优先级的task在同一个qeueue当中。所以可以认为，rt的queue是一个二维数组，横向的表示相同的优先级的task,纵向的表示按照优先级排序。
//上面的代码并只判断prio 小于prev_prio的情况，并没有考虑大于等于的情况，难道是说，无论如何都回去更新么？那这就不是记录最高优先级了，而是记录最近入队的优先级了。其实在inc_rt_prio_smp函数中有去判断的。

	inc_rt_prio_smp(rt_rq, prio, prev_prio);
}

static void
inc_rt_prio_smp(struct rt_rq *rt_rq, int prio, int prev_prio)
{
	struct rq *rq = rq_of_rt_rq(rt_rq);

#ifdef CONFIG_RT_GROUP_SCHED
	/*
	 * Change rq's cpupri only if rt_rq is the top queue.
	 */
	if (&rq->rt != rt_rq)
		return;
#endif
	if (rq->online && prio < prev_prio) //这个条件保证了，只有最高优先级更新的时候才去改变cpupri的值。从而也就保证了cpupri中记录的是各cpu中优先级最高的优先级。
		cpupri_set(&rq->rd->cpupri, rq->cpu, prio);
}

/* Convert between a 140 based task->prio, and our 102 based cpupri */
static int convert_prio(int prio)
{
	int cpupri;

	if (prio == CPUPRI_INVALID)
		cpupri = CPUPRI_INVALID;
	else if (prio == MAX_PRIO)
		cpupri = CPUPRI_IDLE;
	else if (prio >= MAX_RT_PRIO)
		cpupri = CPUPRI_NORMAL;
	else
		cpupri = MAX_RT_PRIO - prio + 1;

	return cpupri;
}

#define CPUPRI_INVALID		-1
#define CPUPRI_IDLE		 0
#define CPUPRI_NORMAL		 1

/**
 * cpupri_find - find the best (lowest-pri) CPU in the system
 * @cp: The cpupri context
 * @p: The task
 * @lowest_mask: A mask to fill in with selected CPUs (or NULL)
 *
 * Note: This function returns the recommended CPUs as calculated during the
 * current invocation.  By the time the call returns, the CPUs may have in
 * fact changed priorities any number of times.  While not ideal, it is not
 * an issue of correctness since the normal rebalancer logic will correct
 * any discrepancies created by racing against the uncertainty of the current
 * priority configuration.
 *
 * Return: (int)bool - CPUs were found
 */
//根据上面的英文注释，这个函数应该是找目前的cpu优先级比当前task优先级还低的cpu
int cpupri_find(struct cpupri *cp, struct task_struct *p,
		struct cpumask *lowest_mask)
{
	int idx = 0;
	int task_pri = convert_prio(p->prio);//这个函数对于理解下面的内容是关键。这个函数将原本rt task 优先级数值越低优先级越高的顺序给反过来了，优先级越低，值越小。

	BUG_ON(task_pri >= CPUPRI_NR_PRIORITIES);

	for (idx = 0; idx < task_pri; idx++) {//按优先级从低到高的遍历。
		struct cpupri_vec *vec  = &cp->pri_to_cpu[idx];
		int skip = 0;

		if (!atomic_read(&(vec)->count)) //如果count等于0表示还没有一个cpu的最高优先级是这个值。
			skip = 1;
		/*
		 * When looking at the vector, we need to read the counter,
		 * do a memory barrier, then read the mask.
		 *
		 * Note: This is still all racey, but we can deal with it.
		 *  Ideally, we only want to look at masks that are set.
		 *
		 *  If a mask is not set, then the only thing wrong is that we
		 *  did a little more work than necessary.
		 *
		 *  If we read a zero count but the mask is set, because of the
		 *  memory barriers, that can only happen when the highest prio
		 *  task for a run queue has left the run queue, in which case,
		 *  it will be followed by a pull. If the task we are processing
		 *  fails to find a proper place to go, that pull request will
		 *  pull this task if the run queue is running at a lower
		 *  priority.
		 */
		smp_rmb();

		/* Need to do the rmb for every iteration */
		if (skip)
			continue;

		if (cpumask_any_and(p->cpus_ptr, vec->mask) >= nr_cpu_ids)
			continue;

		if (lowest_mask) {
			cpumask_and(lowest_mask, p->cpus_ptr, vec->mask);

			/*
			 * We have to ensure that we have at least one bit
			 * still set in the array, since the map could have
			 * been concurrently emptied between the first and
			 * second reads of vec->mask.  If we hit this
			 * condition, simply act as though we never hit this
			 * priority level and continue on.
			 */
			if (cpumask_any(lowest_mask) >= nr_cpu_ids)
				continue;
		}

		return 1; //所以只要找到一个最低优先级且满足counter不为0，就返回。
	}

	return 0;
}

如上面的代码所示，idle位于0元素，而normal task位于1元素，而后才是按照rt优先级排序的元素。

实际上由于只有rt调度器当中才会调用cpupri_set。所以，对于idle与normal来讲都是位于1元素位置，如下所示：
 

void init_rt_rq(struct rt_rq *rt_rq)
{
	struct rt_prio_array *array;
	int i;

	array = &rt_rq->active;
	for (i = 0; i < MAX_RT_PRIO; i++) {
		INIT_LIST_HEAD(array->queue + i);
		__clear_bit(i, array->bitmap);
	}
	/* delimiter for bitsearch: */
	__set_bit(MAX_RT_PRIO, array->bitmap);

#if defined CONFIG_SMP
	rt_rq->highest_prio.curr = MAX_RT_PRIO;
	rt_rq->highest_prio.next = MAX_RT_PRIO;

由于rt_rq->highest_pri.curr的初始值为MAX_RT_PRIO.所以，在rq online的时候：
此时没有rt task，因此highest_prio.curr的值为100

/* Assumes rq->lock is held */
static void rq_online_rt(struct rq *rq)
{
	if (rq->rt.overloaded)
		rt_set_overload(rq);

	__enable_runtime(rq);

	cpupri_set(&rq->rd->cpupri, rq->cpu, rq->rt.highest_prio.curr);
}

根据前面的convert_prio函数的实现，对应的是Normal值。